Property HybridArbitration
- Namespace
- LMKit.Document.Conversion
- Assembly
- LM-Kit.NET.dll
HybridArbitration
Under Hybrid, weighs the vision model's
reading of a page that also carries a text layer against the text tier's reading of
the same page, and keeps the page's own text when the model's reading dropped or
shuffled what the page states or restructured its headings. Defaults to true.
public bool HybridArbitration { get; set; }Property Value
Examples
// Take the vision model's reading of every image page unchecked.
var options = new DocumentToMarkdownOptions { HybridArbitration = false };Remarks
The text layer is exact for the words it holds; the model adds what the images hold.
The check is graded: trust in the model's reading rises with the share of the page's
words it reproduces in order and falls with heading disagreement, and the reading is
kept above a trust threshold. A page kept from its text layer reports
TextExtraction as its strategy and states
the judgement in Warning. Set to
false to take every model reading as it comes.